INSTRUCTIONS

1. Simulate Returns  
Run the following MATLAB scripts to simulate excess returns using both linear and nonlinear data-generating processes.
    - DGP_linear.m  
    - DGP_nonlinear.m


2. Validate Forecast Confidence Intervals  
Run the following scripts to validate the forecast confidence intervals proposed in the paper:  
    - lasso_prediction_bands_simulations.ipynb  
    - nn_prediction_bands_simulations.ipynb

3. Real Data Forecast Intervals  
The codes for estimating forecast intervals using real data are also included.  

    - lewellen_real_data.ipynb
    - lasso_real_data.ipynb  
    - neuralnet3_real_data.ipynb


    - Note: The codes above require CRSP price data, which is not publicly available.  
    - However, the codes are modular and can be easily adapted to any predictor dataset.  
    - Please email me if you would like a sample of how the predictor dataset should be structured.

---

© Rohit Allena. All rights reserved.  
If you use this code in your work, please cite:  
**"Confident Risk Premiums using Machine Learning Uncertainties" by Rohit Allena**
